Countability in world Englishes...context of TESOL and the descriptive contexts of world Englishes and English as a Lingua Franca. We advocate the use of web-based corpora to investigate

World Englishes, Vol. 32, No. 1, pp. 1–22, 2013. 0883-2919

Countability in world Englishes

CHRISTOPHER J. HALL,∗ DANIEL SCHMIDTKE† AND JAMIE VICKERS‡

ABSTRACT: In this study we explored variation in the countability of nouns in Outer Circle, ExpandingCircle and lingua franca Englishes, a phenomenon which is frequently cited as a marker of Inner Circlenorms in TESOL and of endonormative and emerging varieties in the Outer and Expanding Circles. Weinspected a set of mass nouns like information and equipment in the VOICE corpus and websites fromOuter and Expanding Circle country domains. We also evaluated potential causes of variation, investigatingdifferences between Outer and Expanding Circles and the contribution of substrate influence. Our data shownotable and widespread countable use of nouns that are generally non-count in Inner Circle Englishes, butsuch usage is highly infrequent overall. There appears to be greater variation in the Outer than the ExpandingCircle, but little evidence of a determining role for substrate influence. We conclude that the prominencegiven to countability as a marker of ‘nativeness’ and ‘non-nativeness’ is unhelpful, in both the prescriptivecontext of TESOL and the descriptive contexts of world Englishes and English as a Lingua Franca.We advocate the use of web-based corpora to investigate lexico-grammatical variation in lingua francausage and to reveal the ‘plurilithic’ nature of English.

INTRODUCTION

In the speech and writing of some users of Outer and Expanding Circle Englishes, massnouns like equipment, homework, advice, and furniture are attested with countable gram-mar, including plural morphology (advices, furnitures) and number-specific determiners(an advice, several furnitures). This so-called ‘countable usage of non-count nouns’ is amuch-cited lexico-grammatical feature in research on the resources of Englishes beyondthe Inner Circle and their deployment within and between national varieties (Platt, Weberand Ho 1984; Williams 1987; Ahulu 1998; McArthur 2002; Seidlhofer 2005; Jenkins2006; Schmied 2006; Bjorkman 2008; Crystal 2008; Kachru and Smith 2008; Mesthrieand Bhatt 2008; Kirkpatrick and Deterding 2011; Schneider 2011). It also receives a greatdeal of attention in materials for the learning, teaching, and testing of Inner Circle norms(e.g. MacKay 2002; Schoenberg 2005; Yates 2006; DeCapua 2008; Brook-Hart 2009; cf.also Quirk 1990: 8). In this study we attempt first to quantify variation in the ways thatusers deploy mass nouns, and then to evaluate potential explanations for this variation,investigating the difference between Outer and Expanding Circle samples and the role ofcross-linguistic influence from substrate languages.1

In the first decade of the twenty-first century, much scholarly debate on Englishesrevolved around the pluricentricity of ‘the’ English language on the one hand, and thepragmatics of English ‘languaging’ on the other, with the World Englishes paradigm

∗York St John University, Centre for Languages and Linguistics, Lord Mayor’s Walk, York, YO31 7EX, UK. E-mail:[email protected]

†McMaster University, Togo Salmon Hall 605, 1280 Main Street West, Hamilton, Ontario L8S 4M2, Canada. E-mail:[email protected]

‡SARA International Language School, 672–12, Jeonha-dong, Dong-gu, Ulsan City, South Korea. E-mail: [email protected]

C© 2013 Blackwell Publishing Ltd

2 Christopher J. Hall, Daniel Schmidtke and Jamie Vickers

concentrating on the former, and the newer field of English as a Lingua Franca (ELF)studies on the latter. The purpose and design of the research reported here are informedby thinking from both orientations. Like Kachruvian world Englishes, we explore thediversity of English around the world and continue to challenge the ‘apotheosis of thenative speaker’ (Rajagopalan 1997: 229). And like ELF studies we move beyond nationalvarieties to examine the nature of English in its most common contexts of use outside theInner Circle: as a variable and dynamic set of linguistic resources exploited as they areneeded in communication between speakers with different first languages.

Countability is an interesting feature to explore in this regard, because the countableuse of some English nouns is perceived and presented as a salient marker of Outer andExpanding Circle usage, and yet it has no obvious effect on communicative effectiveness(e.g. Bjorkman 2008). In line with Widdowson’s (1994: 381) insight about the status ofgrammatical features as ‘social markers’, we can recognize countability as a purely formalshibboleth of the contested native vs. non-native speaker dichotomy (cf. Quirk 1990: 8;McKay 2002: 127; Higgins 2003: 640; Mollin 2007: 180–1). For example, Crystal (2008: 6)invokes countability as social marker in the following passage, part of a brief speculationon some of the future forms of English:

Some people might think [the countable use of ‘mass’ nouns] ‘un-English’, but in fact informations wasin English once: an information and informations can be traced back to Middle English, and are foundin Chaucer, Shakespeare, Swift, and many other authors. It may only be a matter of time before they areback.

We would argue that they have never gone away. English has never been a monolithicsystem of fixed forms, and multilingual or L2 users have been a part of its story fromthe very beginning. The notion of ‘plurilithic’ English (Makoni and Pennycook 2008)encompasses the social practices (and mental resources: Hall, forthcoming) of millions ofindividual users around the globe, who are all either multidialectal or multilingual, andwhose individual versions of the language vary in form and function. A major componentof scholarly efforts to problematize the ontological status of fixed varieties of English, andso reveal the plurilithic nature of the language, must, in our estimation, include analysesof the degree to which its forms vary (or conform) across users and uses. For applied lin-guists, this enterprise also has a practical purpose, informing decisions about the inclusionof worthwhile and attainable ‘targets’ in pedagogical materials and curricula. Given itsprominence in the literature and yet its low communicative significance, countability is asuitable candidate for such analysis.

Our focus is on variation in the lexico-grammatical forms deployed and experienced byindividuals in lingua franca scenarios, rather than within (more or less institutionalized)national varieties, as represented, for example, in the five Outer Circle corpora of theInternational Corpus of English (ICE 2010). Users of English in lingua franca scenariosare extremely diverse in terms of the circumstances in which their languages have devel-oped, the breadth and depth of the functional repertoires they control, the extent of theirmultilingualism, and the kinds of linguistic practices they typically engage in. The use ofthe binary categories Outer and Expanding Circle Englishes is, therefore, an oversimpli-fication, but a useful one if we wish to characterize the broad distinction between userswho have been exposed to English, and have developed their knowledge of it, in primarilysecond language or foreign language contexts. Although we make use of the distinction


Countability in world Englishes 3

here, what characterizes the phenomenon under study is its concentration in the speechand writing of users beyond the Inner Circle, where lingua franca usage predominatesand native-speaker norms are not the inevitable outcomes of acquisition. We used twovery different sources of authentic data to tap this reserve of international usage: first, weexamined the VOICE corpus of oral interactions constituting English as a Lingua Franca(VOICE 2009); second, we sampled webpages in English in the internet domains of 14countries from outside the Inner Circle, using the Google search engine.

The paper is structured as follows. In the first section, we sketch the way countabilityworks for speakers of English and other languages, describing some of the basic groundplans which could surface in English through substrate influence. We then address thenature and causes of variability in usage by users from beyond the Inner Circle, emphasizingthe disproportionate attention paid to the phenomenon by general and applied linguists,as well as TESOL professionals. The second section reports data, first from the VOICEcorpus and more substantially from Internet snapshots; we then discuss potential causesof variability, assessing the evidence for processes of innovation and/or nativization. Thearticle concludes with a discussion of the data framed by our understanding of the plurilithicnature of English and the consequences of this for establishing and teaching language‘norms’.

COUNTABILITY IN ENGLISHES AND OTHER LANGUAGES

In this section, we describe the phenomenon of countability as it is expressed in Englishand other languages. The kind of concepts routinely expressed by languages as nouns(henceforth ‘nominalizable concepts’) may be mentally categorized on various dimen-sions, including whether they are inherently atomic (particularized, individuated wholes)or inherently mass (non-particularized collectives). Solid, bounded objects (e.g. ‘leaf’ or‘chair’) tend to be categorized as atomic, whereas nonsolid, dispersed (classes of) objectsor ideas are more likely to be categorized as mass. Mass concepts include: (a) fluids andother indefinite collections of atoms (e.g. ‘water’ or ‘gravel’); (b) (kinds of) substance (e.g.‘meat’ or ‘wood’); (c) generic concepts (e.g. ‘hair’ or ‘lightning’); and (d) qualities (e.g.‘redness’ or ‘honesty’). The dichotomy is essentially one of ‘individuation’: nominalizableconcepts are either taken as individuals which can be counted and therefore linguisticallyexpressed as plural, or as groups or masses which would not normally be counted becausethey are unindividuated (inherently plural) to begin with.

For the most part, the way our minds categorize the ‘individuability’ of nominalizableconcepts is common across all human beings, independently of the language(s) we acquirein infancy (cf. Xu 2007; Barner, Inagaki and Li 2009). Furthermore, the conceptualcategories we construct internally are externalized by different language systems usinga predictable and quite well understood set of grammatical devices (Chierchia 1998;Aikhenvald 2003). But the semantic mapping from concept to form is not uniform acrossall language systems, and the grammatical devices used for this mapping are not universal.Language systems exploit two basic designs for expressing individuation:

1. countability via number morphology (e.g. page vs. pages); and2. non-countability via classifiers (e.g. paper vs. pieces of paper).

English, Swahili and Sinhala are languages which (mostly) rely on countable gram-mar; Mandarin Chinese, Japanese, and Filipino use non-countable grammar. No language



system uses the countability option exclusively, whereas many language systems employnoncountability, and in fact it appears to be the unmarked option cross-linguistically (Allan1980; Chierchia 1998).

For users of Inner Circle (henceforth IC) Englishes, most nominalizable concepts (inde-pendently of individuability) are expressed as count nouns and are morphologically markedas singular or plural, with determiner modification required in some contexts. There is,however, a group of nouns which have limitations on their accessibility to count grammar.The group is a tiny subset of the set of all English nouns, but its membership is largerthan similar groups in other non-classifier languages. Spanish, for example, has singularand plural forms for advice and furniture, and Sinhala expresses many of the same nomi-nalizable concepts, including advice and furniture, only in plural form (so-called pluraliatantum, like entrails in English). Although countability undeniably plays a significant rolein determiner selection in English (e.g. this information vs. these informations), the smallclass of nouns which are non-count in IC Englishes actually constitutes a very negligi-ble element in the lexico-grammatical paraphernalia of the language. Indeed, the sectionsdedicated to countability in Quirk, Greenbaum, Leech and Svartvik’s (1985) painstakingdescription of IC (‘Standard’ British) grammar take up fewer than seven out of over 1,600pages (less than half a per cent). Moreover, the group has leaky borders. According toAllan’s seminal (1980) study, countability in (IC) English(es) is best understood as a con-tinuum on which nouns may be placed according to their ‘countability preferences’, ratherthan in terms of a binary [+count] or [-count] feature marked on each noun.

Mass nouns in the Englishes of multilingual learners and users

Despite its knotty and negligible status, the count/non-count distinction has a strikinglyhigh profile in descriptions of the Englishes taught to and used by native speakers ofother languages. It figures especially prominently in mainstream English teaching andtesting materials and practices. Among textbooks, for example, it features very early onin Pearson’s Focus on Grammar (Basic) (Schoenberg 2005). It is listed as the locus of oneof the top ten EFL ‘mistakes’ in Brook-Hart’s (2009) Learning from Common Mistakesbooklet, based on the Cambridge Learner Corpus of exam papers (and billed as part of its‘Real English Guarantee’). The popular www.onestopenglish.com website provided by thepublisher Macmillan supplies five intermediate level lesson plans for it (out of 74 overall)– more than for word order or modals – and it is listed as the most popular grammaritem in a sidebar. Yates’s (2006) English Vocabulary for Beginning ESL Learners actuallystarts out with a very challenging introduction to count and mass nouns (see below).For the professional development of teachers, too, it is seen as worthy of considerableattention: DeCapua’s (2008) Grammar for Teachers, for example, dedicates 12 pages to it,and Jacobs’ (1995) English Syntax. A Grammar for English Language Professionals hasalmost a whole chapter on the subject.

In Quirk’s (1990: 8) impassioned defence of native-speakerism, his notorious referenceto ‘half-baked quackery’ was levelled against an English teacher who questioned whyhis students’ use of several informations should necessarily be corrected. The need tovigorously teach IC norms for such nouns is taken for granted in most ESL and EFLmaterials and curricula, despite the marginality of the distinction and its considerableintractability. The complexity revealed by Allan (1980) and apparent too in Quirk et al.’s(1985) publicly authoritative treatment does not appear to have daunted practitioners, nor



prevented many of them from assuming that learners need early mastery of it. An indicationof the heightened importance associated with it in resources for learners is the attentionit is allocated in grammars for teachers: Carter and McCarthy, in their (2006) grammarfor learners and teachers, dedicate over twice as much space to countability as Quirk etal. (1985). But the pedagogical response in EFL and ESL teaching materials on the topicis not distinguished by its clarity and accessibility to learners. Consider, for example, thefollowing passage from the very first page of Yates’ English Vocabulary for BeginningESL Learners (2006: 1), which follows an initial introduction to the count/non-countdistinction.

Other nouns cannot be counted – air, wind, and pollution, for example. They have no plural forms, areused with singular verbs, and are called “noncount” nouns. But noncount nouns can also be things thatwe can count! First, there are those that it would take a lifetime to count, so we call them by a moregeneral noncount noun, such as hair, sugar, or flour. And then there are those that we categorize intogeneral groups that are named by noncount nouns, such as furniture, mail, silverware, and china. Ofcourse we can count chairs, tables or beds, but the general category furniture is never made plural. Thenoncount noun mail includes the letters and cards that we can count. English has a lot of these words.

The explanation mixes statements of ‘fact’ (‘cannot be counted’, ‘is never made plural’),with more homely descriptions of what ‘we’ (as IC owners of English) do (‘we call themby a more general noncount noun’; ‘we can count chairs’; etc.). Students beginning thebook must feel some trepidation about the learning journey ahead when confronted soearly in the process with this unequivocal, but somewhat random, characterization of ICEnglish (cf. Kachru and Smith 2008: 90).

Other TESOL professionals simplify the challenge, but still recognize that it is a sig-nificant one for learners. The following is taken from a live chat transcript of a ‘grammarsurgery’ offered by the BBC World Service Learning English website (BBC World Service2004):

[Learner]: How can we know which noun is countable or uncountable? Is there a rule?

[‘Grammar Masterclass’ Expert]: Most nouns are either countable or uncountable. It is logical which iswhich. For example, clearly we can count books so the word ‘book’ is countable. On the other hand,we can’t count ‘water’ so the word ‘water’ is uncountable. Unfortunately, though, there are a few wordswhere this distinction isn’t so clear. ‘Furniture’ is a good example of this. We can’t say ‘I bought somefurnitures’ because ‘furniture’ is uncountable. However, lots of learners of English think (rightly!) thatthis is rather illogical.

But whether it is logical or not, it is part of the monolithic system that most learners areexpected to internalize, even when there is no evidence that it contributes to communicativeeffectiveness. Jenkins (2006: 44) signals the injustice of assuming a monolithic account ofIC countability for learners:

[A] candidate in an ELT speaking exam would be rewarded for their knowledge of ‘real’ English if theywere to say ‘three teas’ or ‘two coffees’ instead of ‘three cups of tea’ or ‘two cups of coffee.’ On theother hand, if they extended this use of uncountable nouns to ‘wine’ and referred to ‘two wines’ insteadof ‘two glasses of wine’ they could be penalized for lack of competence with the countable/uncountabledistinction.



In confirmation of this, Lowenberg (1992: 116) discusses items in the Test of Englishfor International Communication in which test-takers must indicate counted ‘non-count’nouns (like resistances and equipments) as ungrammatical, although they ‘may well beacceptable to educated speakers of Malaysian or other non-native varieties of StandardEnglish’. A similar point is made by Bamgbos.e (1998: 4) for West African English and byShim (1999) for codified Korean English.

In studies of the Englishes of users in the Outer Circle (henceforth OC), the so-called‘countable usage of non-count nouns’ is widely cited as a distinctive feature, especiallyin African and Asian varieties (e.g. McArthur 2002; Schmied 2006; Mesthrie and Bhatt2008; Schneider 2011). Platt et al. (1984: 50) observed that ‘nouns are sometimes markedfor plural in the New Englishes where they would not be marked in the establishedvarieties of English’, and they attribute this to the ‘reclassification’ of uncountable nounsas countable. Williams (1987: 171) states that ‘this [count/non-count] distinction frequentlydoes not follow the rules of NS varieties’. For Schneider (2011: 204), the ‘pluralizationof mass nouns’ is listed as a ‘fairly common’ characteristic of New Englishes, whichis ‘widespread in Africa and Asia’. Mesthrie and Bhatt (2008: 53) state that ‘[a]lmostevery study of individual WE varieties in Africa and Asia reports frequent exampleslike furnitures, equipments, staffs, fruits, accommodations, and less common ones likeoffsprings, underwears, paraphenalias [sic], etc.’ Kirkpatrick and Deterding (2011: 378)also claim that the ‘occurrence of furnitures and similar words’ is ‘widespread’ in OCEnglishes.

Some scholars go slightly further. Ahulu (1998: 19–20), in a paper focusing on numbermarking on nouns and in verbal concord, highlights countability as ‘the most notable andregularly cited’ among ‘major area[s] of divergence between British Standard Englishand the English written in postcolonial countries’. Schmied (2006: 198) claims that ‘EastAfrican usage basically ignores the grammatical distinction of count vs. non-count nouns[.]’ Kachru and Smith (2008: 106) refer to ‘the extensive use of collective nouns ascountable in almost all OC varieties’.

For Expanding Circle (henceforth EC) Englishes and English as a Lingua Franca (ELF),in which Englishes from all three circles might be represented, the phenomenon hasalso been noted as a salient characteristic. Seidlhofer’s (2005: R92) earlier code-orientedwork on ELF includes the ‘plural use of mass nouns’ as one of eight lexico-grammaticaltendencies. Jenkins (2006: 44) states that forms such as a staff and four furnitures ‘areused by many speakers of the Expanding Circle’ and that processes of regularization inELF are likely to lead to features such as ‘the countable use of nouns that in ENL areconsidered uncountable (e.g. informations, advices)’ (Jenkins 2009: 201). Shim (1999)cites the use of ‘non-count nouns as count nouns’ as one of a dozen morpho-syntacticfeatures which distinguish Korean English from US English. Mollin (2007) includes itemssuch as informations and equipments in an acceptability test for users of Euro-English. AndBjorkman’s (2008) study of ELF usage in Swedish tertiary education identified ‘incorrectplural forms/countability’ as one of three morphological features.

In sum, the so-called ‘countable usage of non-count nouns’ is portrayed as one of themost salient characteristics of Englishes beyond the IC. On the one hand it is viewed as anerror to be prevented or corrected as early as possible in the learning and teaching of ICEnglish, and on the other is presented as a common feature of new endonormative varietiesof English in the OC, of emerging varieties in the EC, and of the formal resources deployed



in ELF interaction. Before examining samples of English from beyond the IC to assessthe prevalence of the phenomenon, we address the question of its causes, and in so doingbegin to question its status as non-IC variety marker.

Variation in countability and its possible causes

As we have noted, scholars (notably Allan 1980) have demonstrated that countabilitydoes not work in a binary fashion in the grammars of even monolingual IC users. This facthas been acknowledged and highlighted by researchers on world Englishes. Sey (1973),who documented the phenomenon in Ghanaian English, observed:

There appears to be (a) no consistent semantic relationship between countable and uncountable uses ofnouns [in IC Englishes], nor (b) any clearly discernible motivation for using some normally uncountablenouns in countable functions but not others. (cited in Bokamba 1992: 131)

Schmied (2006: 198) echoes this observation, stating that the grammatical distinction‘does not always correspond to the semantic distinction’ and that ‘in [IC Englishes] somenon-countables may occur in the plural in special meanings (e.g. works)’, concludingthat ‘thus differences are often a question of interpretation and frequency’. This lackof transparent semantic mapping and the absence of a simple binary count/non-countdistinction in IC Englishes are at the heart of Allan’s (1980) conclusion that countability isa continuum upon which each noun’s place must be assessed by grammaticality judgementson an item-by-item basis.

If the countability features that IC users construct for nouns in their mental lexicons andgrammars is, to repeat Schmied’s words, ‘a question of interpretation and frequency’, thenwe should not be surprised to find variability also between users and learners in the OC andEC, given that their mental lexicons and grammars are ultimately rooted (either historicallyor pedagogically) in IC usage. Mesthrie and Bhatt (2008, 53–4), signal the interestingpossibility that some OC users might maintain countable forms that were present in the ICEnglishes to which speakers were exposed in the colonial past, but that are now non-countaccording to IC norms. In a corpus of Settler English from early nineteenth century SouthAfrica, Mesthrie and West (1995) found, among other variable structures, ‘plural endingsfor non-count nouns like progresses, evidences, sufferings’ (cited in Mesthrie and Bhatt2008: 191). Further to Crystal’s (2008) observation, cited earlier, that informations has along history in IC English, Toyota (2009) suggests that the count/non-count distinction isa relatively recent development in the history of English, and indeed that Old English andEarly Middle English were essentially classifier languages. In the light of this intrinsicvariability, we recommend dispensing with the ‘native-speakerist’ nomenclature of countnoun vs. non-count noun and their implication of a monolithic binary typology. Here weuse the more neutral term mass noun for that subset of nouns which for IC users are(normally) non-count in the relevant grammatical contexts but which for OC and EC usersmay not be – the group of nouns that Allan (1980: 560) calls ‘true uncountables’.

The extent to which the idiosyncrasy of IC usage has influenced the usage of massnouns in the Englishes of other regions remains to be seen. But it is almost certainly onlyone factor of several at play. Platt et al. (1984: 51–2) include it along with three otherfactors influencing OC users’ ‘reclassification’ of non-count nouns as countable (cf. alsoMesthrie and Bhatt 2008: 53–4):



� Countable semantics: Some mass nouns can be taken to refer to ‘separate items, sothat furniture means an item of furniture’.

� Morphological resemblance: Some mass nouns are morphologically related to countnouns (e.g. jewellery – jewels).

� Substrate influence: Some mass nouns are treated by ‘background languages’ (sub-strates) as countable (although they note that the tendency is not absolute).

A fifth factor is one that applies to a large number of lexico-grammatical (and other)processes in English: the tendency towards simplification, including regularization, whichis attested in both first and second language acquisition (cf. Slobin 1973; Williams 1987).This factor would lead learners to map all nominalizable concepts onto countable nouns,without exception. Of these factors, those related to semantics and substrate are of particularinterest, because they will vary in their effects across nouns and speaker groups, and so aremore likely to reveal the processes of innovation and nativization which are hypothesizedto underlie the emergence of new norms in specific contexts (Bamgbos.e 1998). We exploretheir role in our own data at the end of the next section.

ASSESSING THE SCOPE OF THE PHENOMENON

There are almost no data on the extent of variability in the grammatical deployment ofmass nouns. Given the importance attached to countability as an indicator of normativeuse, both in English teaching guides and scholarly descriptions of World Englishes, wedecided to assess just how extensive the countable usage of mass nouns is in the authenticpractice of users in lingua franca contexts. We began our search with the Vienna OxfordInternational Corpus of English (VOICE 2009) but, as reported below, the data scarcity ofVOICE led us to conduct an expanded search using snapshots of the world wide web, andit is this second source that provides most of the data reported in the following sections.

Countability in ELF

Mollin (2007) is the only study we are aware of which attempts to quantify the countableusage of mass nouns outside the IC. As part of an investigation into the status of Euro-English, representing mostly EC speakers, she constructed a 400,000-word corpus ofEnglish usage in non-Anglophone Europe, composed of about 60 per cent transcriptionsof spoken data from European Commission public discussions, speeches, etc., and 40 percent written data from chat forums and online discussion groups, with over 900 speakersrepresented overall. She found little evidence for widespread countable usage of massnouns. She calculated proportions of countable uses for around 40 mass nouns, including27 mentioned in Swan (1995) and 13 pluralia tantum (Mollin, personal communication).The noun with the highest proportion was bread, at 18.75 per cent, followed by adviceat 11.86 per cent and evidence at 9.68 per cent. Mollin (2007) points out that not onlyare these proportions low, but also that countably used mass nouns are unrepresentativeof the full range of speakers and texts in the corpus. She notes, for example, ‘[a]ll threeinstances of bread used as a countable noun stem from the same Finnish speaker in thesame text, and two of the three countable uses of evidence come even from the samesentence’ (Mollin 2007: 179). This pattern is repeated for other nouns, and overall, theaverage rate of countable usage of the mass nouns examined was 2.54 per cent.



Mollin’s study has been criticized on both methodological and conceptual grounds bySeidlhofer (2009). One problem she identifies regards the representativity of the genressampled, for example, the emphasis on public debate, which is ‘very unlikely to support anatural vernacular way of speaking’ (Seidlhofer 2009: 46). For a more ‘natural vernacular’sample we chose Seidlhofer’s VOICE corpus to begin our quantitative assessment ofmass nouns in non-IC usage. Conceptually, Mollin is criticized by Seidlhofer for adoptingan overly ‘formalist’ orientation, which prevents her from problematizing the traditionalENL/EFL/ESL typology. Like Mollin, we are unapologetically ‘feature spotting’ here, butthis does not represent for us a blinkered ‘fixation on form’, as we hope to show in ourconcluding discussion.

VOICE contains just over one million words from speech events involving 694 L2speakers (and 57 L1 speakers) in mostly professional contexts. Fifty one languages arerepresented, although 87 per cent of them are Indo-European and just over a third of thewords in the corpus are spoken by L1 speakers of German (25%) and Dutch (10.5%)(VOICE 2009). To generate a sample set of mass nouns to search for, we compiled a (notexhaustive) master list of 183 potential items on the basis of listings of nouns labelled as‘mass’, ‘uncountable’ or ‘non-count’ in grammar books, websites, and scholarly articles.For each master list item, we conducted a search for singular and plural stems and recordedcountable usage when the noun was situated in one of the following contexts (examplesare from VOICE):

1. occurrence with plural -s, for example:

(a) [ . . . ] and help them for their homeworks [ . . . ](b) [ . . . ] the way of reaching informations [ . . . ]

2. preceding indefinite article or numeral one, for example:

(a) [ . . . ] just one luggage [ . . . ](b) [ . . . ] shall I just buy a milk quickly [ . . . ]

3. preceding ordinal number, for example:

(a) [ . . . ] the third advice again comes from our xx [ . . . ](b) [ . . . ] our fo[u]rth advice will be to monitor international currency flows [ . . . ]

Of the 111 types occurring in VOICE, only 19 occurred with instances of countablegrammar in a total of 52 tokens used by 45 speakers. Table 1 shows a summary of countableusage of mass nouns in VOICE.

Among mass nouns frequently cited in the literature that had zero countable usage inVOICE are staff, baggage, furniture, evidence and equipment. Inspection of contexts of useindicate that of the 111 master list noun forms attested, 31 appeared in contexts in whichnon-countable grammar would have been inappropriate (because of polysemy in which anon-mass sense was being expressed), resulting in 80 (111 minus 31) plausible mass nountokens to count. This means that, in effect, fewer than 25 per cent of the types investigatedwere used countably (19 of 80), and that of these types, countable tokens accounted foraround 2.9 per cent of all occurrences. Users of mass nouns with countable grammar wereL1 speakers of 15 different languages, most of them European.

We may conclude from this that countable usage of mass nouns is uncommon in linguafranca interaction between mostly European EC speakers. Our finding (2.9%) is of the



Table 1. Mass nouns (n = 19) used countably at least once in VOICE, ordered by number of countable uses

Countable usesTokens Noncountable uses Speakers

Mass noun n n n (%) n

advice 30 23 7 23.3 5information 407 400 7 1.7 6money 473 467 6 1.3 4knowledge 201 196 5 2.5 4employment 88 84 4 4.5 4research 136 132 4 2.9 4paper 27 24 3 11.1 3traffic 18 15 3 16.7 3luggage 6 4 2 33.3 2violence 24 22 2 8.3 1harm 2 1 1 50.0 1homework 14 13 1 7.1 1lack 78 77 1 1.3 1luck 30 29 1 3.3 1milk 35 34 1 2.9 1permission 10 9 1 10.0 1public 70 69 1 1.4 1reading 9 8 1 11.1 1stuff 152 151 1 0.7 1Total 1810 1758 52 2.9 45

same order as the 2.5 per cent found in Mollin’s study. But despite the low rate of countableusage, it is still notably above the norm for IC usage, which, according to descriptivelinguists, is for non-count grammar across all mass nouns in the contexts inspected. Toconfirm this from actual user data, we obtained from the British National Corpus (BNC)the frequencies of plural usage for 16 of the 19 mass nouns occurring in VOICE (weomitted paper and reading because of high plural usage in non-mass semantic contexts,and permission, because of its specialized plural usage in publishing). The overall rate ofcountable usage for this sample in the BNC was just above zero, at 0.3 per cent.2 A t-testrevealed that the difference between the VOICE and BNC rates was highly significant(t(15) = 2.779, p < 0.01).

It would, of course, be reckless to make generalizations about countability in non-ICusers on the basis of these data alone. Because of the heightened degree of mobility andinteraction with native speakers that European standards of living and location bring, onemight expect the kind of professional discourse used by the educated Europeans of Mollin’scorpus and much of the VOICE corpus to be more likely to conform to IC norms. Thismeans that the extent of countable usage of mass nouns in lingua franca English may beseriously under-represented by these results. Although corpora like VOICE are rich incontextual detail, ‘data scarcity’ is one of the inevitable consequences, and this appearsto limit their usefulness for frequency studies of lexical particularities like countability.We therefore sought a more data-rich sample of Englishes in lingua franca usage, in orderto maximize the pool of types and tokens we could analyse and so obtain a wider-angledsnapshot of the prevalence of countable mass nouns.



Counting nouns in the world wide web

The world wide web is immensely fertile terrain in which to encounter locally-determined but globally-directed Englishes produced by speakers of international varietiesand in ELF contexts (cf. Gupta 2006). Although only some English language text on OCand EC websites is produced in online interaction (in chat forums, for example), a greatdeal of it is written for and/or read by users who do not share the native language of thewriter. To this extent, it can be claimed to constitute a kind of lingua franca English, evenif most of it is not necessarily ‘crafted ELF’, displaying the kinds of deliberate accommo-dation or interactive negotiation of meaning represented in VOICE. Moreover, the web ascorpus is staggering in its size, dwarfing even the biggest purpose-built language corpus:according to current best estimates, there are over half a billion primarily English-speakingusers of the Web, 42 per cent of all English users world-wide, responsible between them forat least 28 trillion English words (Norvig 2007; Internet World Stats 2010). Indeed, thesestatistics hugely underestimate English usage because they make the grossly simplifyingassumption of global monolingualism, that is, attributing only one ‘official’ language toeach nationality.

Like VOICE, the web contains much ‘natural vernacular’ usage; but unlike VOICE, it isnot a structured, sampled collection of text. This, we argue, makes it a particularly attractivesite to search for variability in the use of mass nouns. Much of the early work using the webas a collection of English text was conducted by researchers in the natural languageprocessing community, who have consistently sought ways to eliminate the ‘noise’ in thesystem that ELF and world Englishes scholarship is keen to reveal. For example, in orderto avoid ‘random noise caused by misspelled words, non-native speakers, pages in otherlanguages, etc.’, Villavicencio (2003: 9) searched the web for verb-particle constructionsin contexts constrained by descriptive accounts of IC norms, rather than simply searchingfor the verb and particle forms alone. And from an even more unequivocally prescriptiviststandpoint, Fletcher writes:

As the lingua franca of the digital frontier, English is both the target and source of contamination:non-Anglophones often translate their web pages into Info-Age pidgin English while fusing creolizedweb English into texts in their native tongue. Similarly, searches for the linguistic examples [sic] canlead to work by learners with imperfect mastery of the language or to baffling machine translations.In many online forums, careless or cryptic language and sloppy spelling prevail. With its frenetic paceof development, the web typically values content creation above perfection and tolerates ill-formedlanguage. (Fletcher 2007: 36).

We chose the web as a database precisely to hear some of this so-called ‘noise’ and‘ill-formed’ language: just like ELF, it is transnational, un-normed and dynamic, a majorsite of Jacquemet’s (2005) ‘transidiomatic practices’. As such it provides an ideal searchspace for an assessment of the scope of countable use of mass nouns in individual Englishesoutside the IC.

We cannot offer a very refined analysis here and, given the numbers of pages sampled,make no attempt at qualitative analysis. A country code on a url is no guarantee ofprovenance, let alone authorship, so provides no information on the nationality of, orlanguage(s) known and/or used by, the author; nor whether the text on the page is writtenby one or many authors. Indeed, just as many pages in the domains of IC countries willcontain text written by OC and EC speakers, so too will there be significant amounts of text



authored by IC speakers on OC and EC country domains. And while the web may be a keylocus of global Englishes in lingua franca usage, in its unmediated and unfiltered state itpresents other serious problems for the investigator, including questionable text authenticity(e.g. ‘spamdexing’ and ‘keyword stuffing’ for commercial purposes), volatile data shifts(webpages added, deleted, modified, and copied), and very limited options for setting searchcriteria (e.g. there is no part of speech tagging). Despite this intractability, snapshots of theweb have been demonstrated to be very effective for assessing the frequency of languagestrings across millions of speakers in multiple genres: research shows that the number of‘hits’ for pages containing particular strings correlates very well with BNC frequenciesfor those strings (Keller and Lapata 2003) and indeed that the web outperforms corporafor natural language processing simulation purposes (Lapata and Keller 2005). Althoughspecialized tools like WebAsCorpus.org and WebCorp can now significantly refine thesearches performed by commercial search engines like Google, Bing and Altavista, at thetime of data collection we determined that Google would be adequate for our purposes,given that what we wanted was a wide-angled, quantitative snapshot of the presence ofpluralized mass nouns on OC and EC Internet pages.

METHOD

We used the advanced settings of the Google search engine to find pluralized massnouns in pages tagged as English in sites hosted by 14 different polities, using theircountry codes. Bergh (2005: 43) has highlighted the recognition by corpus linguists that‘while it is generally difficult to use the Web in its entirety for frequency studies [ . . . ],slices of it in terms of domain-specific searches are more rewarding as they lead to ahigher level of precision[.]’ Renouf (2003: 49–50), for example, suggests (for oppositepurposes to ours) that using the .uk country domain ‘might limit the retrieval of non(English)-native-speaking text and word use’. We limited our searches to specific countrydomains from the Outer and Expanding Circles, with the language domain set to English,precisely to tap into the Englishes of multilingual users in lingua franca scenarios. Searcheswere performed in the English-tagged internet domains of 13 nations plus the Hong KongSpecial Administrative Region. We also performed a search on the .uk domain as anIC baseline. The 14 non-IC domains searched were: Outer Circle – Bangladesh (.bd),Belize (.bz), Hong Kong (.hk), Kenya (.ke), Malta (.mt), Philippines (.ph), Sri Lanka (.lk);Expanding Circle – Angola (.ao), China (.cn), Iceland (.is), Mexico (.mx), South Korea(.kr), Thailand (.th), Yemen (.ye).

The selection was intended to represent a broad sample of English-using domains fromthe OC and EC (seven from each). Given our interest in potentially tracing cross-linguisticinfluence from speakers’ L1(s) and other national languages, we strove to select countriesin which there was a numerically-dominant first language (or language family) with whichIC non-count grammar norms and/or preferences could be contrasted. Most domains had aclearly dominant L1, at over 70 per cent of speakers according to Ethnologue (Lewis 2009).The three major exceptions were: Belize, with Spanish at 35 per cent; the Philippines, withFilipino (including Tagalog) at 46.5 per cent; and Thailand, with Thai and related Tailanguages at 46.6 per cent. Some domains (e.g. Kenya and Angola, Belize and Mexico, SriLanka, and the Philippines) were included because their principal L1s used grammaticalnumber in ways which contrast significantly with IC Englishes. Others (Bangladesh, Hong



Table 2. Country data. Language data from Ethnologue (Lewis, 2009); internet data from Internet WorldStats (2010)a.

Principal L1

Country Language n (%) Typea Internet penetration (%)

Outer Circle Bangladesh Bengali 110m 77 Class 0.4Belize Spanish 0.1m 35 Num 19.5Hong Kong Cantoneseb 6m 91 Class 69.9Kenya Bantu 16.8m 60 Num 8.6Malta Maltese 0.3m 71 Num 59.7Philippines Filipinoc 46.5m 46 Neither 35.1Sri Lanka Sinhala 15.5m 80 Num 5.5

Expanding Circle Angola Bantud 11.9m 99 Num 3.4China Mandarin 840m 69 Class 28.7Iceland Icelandic 0.2m 100 Num 100Mexico Spanish 86m 93 Num 24.8South Korea Korean 42m 100 Class 77.3Thailand Thaie 46.6m 90 Class 24.4Yemen Arabic 15.5m 99 Num 1.8

Notes:aClass = classifier language; Num = number morphology.bFigures from Government of Hong Kong SAR (2007).cIncludes Tagalog.dVarious Bantu languages.eIncludes related Tai languages.

Kong, China, South Korea, the Philippines and Thailand) were included because theirprincipal L1s are classifier languages. We also strove to include countries from all globalregions, with East and South-East Asia particularly well represented because of theirreported widespread countable use of mass nouns in English, as well as their high Internetpenetration and/or large number of webpages. Some domains (Bangladesh, Sri Lanka,Yemen, Angola, Kenya) have very low internet use, and so of course the number of tokensin our database is low for them compared with other domains, and is less representative ofthe nation as a whole than, say, the figures for Iceland or South Korea, where internet useis an element of most or all citizens’ cultural capital. Table 2 summarizes the country data.

The master list of mass nouns was expanded to 215 from the 183 used in the VOICEsearch, and although still inevitably incomplete, provided a sufficiently representative poolof words from which to identify a subset of Allan’s (1980) ‘true uncountables’ to searchfor in Internet domains. Using Allan’s rankings of countability preferences, a subset of41 potential candidates for search was isolated from the expanded master list and eachitem was scrutinized for syntactic (e.g. word class) and semantic (e.g. homographic orpolysemic) ambiguities. Numerous candidate nouns had to be eliminated on these groundsbecause of the likelihood that related forms which were not mass nouns would be includedin the count. For example, syntactically, research is a common verb as well as a noun, andso instances with a suffixed -s may be third person singular verb forms as well as pluralnoun forms. Semantically, paper is a mass noun when it refers to the material or a blanksheet, but is a count noun when it refers to written documents (including newspapers).



The candidate list was thus narrowed down to 25 mass nouns:

1. advice 10. feedback 19. luggage2. applause 11. fun 20. magic3. baggage 12. furniture 21. slang4. cash 13. hardware 22. software5. corruption 14. homework 23. traffic6. dew 15. information 24. underwear7. employment 16. jewellery 25. violence8. equipment 17. knowledge9. evidence 18. luck

For each mass noun, separate searches were conducted in each country domain for thebare stem and the stem plus plural -s, using the ‘exact wording or phrase’ option in theadvanced search pane of the Google search engine.3 The total number of hits per mass nounper domain was recorded and, to ensure representativity, figures lower than 100 tokensper domain were not included in subsequent analysis.4 The proportion of plural tokens ofeach noun type was calculated, yielding an approximate indicator of countable usage of themass noun in the Englishes of the domain searched. Averages were then calculated acrossthe sample set of the domain, as an approximate indicator of overall countable usage ofmass nouns for the domain.

FINDINGS

The average percentage of countable usage across all nouns and country domains was2.45 per cent, a figure which is remarkably consistent with the 2.54 per cent reported inMollin (2007) for Euro-English and the 2.9 per cent we found in the analysis of VOICEdata. This figure confirms the low estimates of the earlier studies, suggesting that countableusage of mass nouns is an infrequent characteristic of Englishes in OC or EC domains. Itis, however, significantly higher in frequency than IC usage, as demonstrated by the verylow rate for the .uk domain and BNC. For the .uk domain, the rate for the sample was justover 0.01 per cent and for the BNC, 0.02 per cent. The difference between these rates andthose for our OC and EC sample are extremely significant according to t-tests performedon the means (t(24) = 5.166, p < 0.00001 for .uk; t(24) = 4.669, p < 0.0001 for BNC).Individual nouns and individual domains (reflecting national varieties) yield some modestbut marked departures from the relatively low rate of OC and EC usage, as can be observedin Figures 1 and 2.

The high rate for the Philippines is due in large part to the behaviour of two nouns:luggage at a solid 87.92 per cent of countable usage (270,000 instances out of 307,100 intotal) and homework at a substantial 41.61 per cent (13,400 instances out of 32,200). Inorder to avoid skewing any estimations of central tendency for the whole class of massnouns in subsequent discussions, we replace the counts for these two words with themean count for the Philippines domain. Aside from these two words, equipment showsconsistently higher rates of plural usage across all domains except Iceland and Malta, withan overall mean of just over 9 per cent. Interestingly, equipments was judged ‘acceptable’by just over 50 per cent of over 400 European academics responding to an acceptabilitytest used as part of Mollin’s study of European ELF (Mollin 2007: 180–1). In sum, our



Figure 1. Countable usage by country domain, lowest to highest.

Figure 2. Countable usage by mass noun, lowest to highest.

figures suggest that, in the speech of (mostly) European ELF users and the written Englishof a diverse range of OC and EC Internet domains, the countable use of mass nounsis: (a) infrequent, even for the most variable nouns; but (b) recurrent compared with ICusage, where it is almost entirely absent; and (c) widespread, attestable across numerousL1 backgrounds and geographical regions.



Investigating causes

In an attempt to identify potential causes of the patterns of occurrence identified, wetested a couple of specific hypotheses regarding language change and language acquisition.These factors have been identified by previous scholars as potentially accounting for thecountable use of mass nouns by OC speakers (e.g. Platt et al.,1984: 51–2; Gramley 2001:119–20) and play a major role in processes of standardization/institutionalization or norm-fixing (cf. Kachru 1992, on ‘deviations’ and Bamgbos.e1998, on ‘innovations’), but hithertohave not, as far as we know, been specifically tested.

First, we considered the status of English in the domains sampled, using the simplifyingbinary distinction second language in the OC vs. foreign language in the EC (Kachru1985). We hypothesized that there would be higher rates of countable usage in the formerthan the latter, given the assumed ‘post-learner’ status and relative norm-independence ofmany Englishes in the OC. As is widely supposed, ‘innovations’ (i.e. non-IC forms andusages) are more likely to occur and be tolerated in endonormative OC contexts than inthe exonormative EC, where IC norms still dominate the TESOL agenda and mindsets ofusers. Second, we explored the role of cross-linguistic influence. Given the lack of pluralmorphemes in classifier languages, we hypothesized that the overall rate of pluralizationfor mass nouns would be higher in classifier language domains than in non-classifierlanguage domains. Within the latter, we hypothesized that nouns translated by pluraliatantum in the substrate language would yield higher rates of pluralization than those withcountable translations, and that these in turn would yield higher rates of pluralization thanthose with non-count translations.

Outer vs. expanding circles

The rate of countable usage of the mass nouns studied in the seven OC country domainswas 3.43 per cent, compared with only 1.01 per cent for the EC. The numbers for the OCare perhaps exaggerated because of the behaviour of the words homework and luggage,especially in the Philippine data. If we replace the percentages for those items with themean, the OC rate falls from 3.43 per cent to 2.48 per cent. But as the graph in Figure 3shows, the tendency for greater countable usage in the OC is maintained, with equipmentthe one major exception, and advice and violence occurring slightly more in EC domains.

This tendency is not an artefact of one or two words or countries. If we concentrate onthe 51 relatively high countability rates of above 5 per cent (representing over 14 per centof all 350 rates calculated), we find that they are distributed across 18 of the 25 mass nounsand occur in all country domains but one (Iceland, in the EC). The OC has 34 individualcountry counts above 5 per cent, compared with only 19 in the EC. Of these ‘high’ rates,12 in the OC are above 10 per cent, compared with only nine in the EC. Overall, thedifference between OC and EC rates is extremely significant statistically (t(24) = 5.204,p < 0.00001).

Cross-linguistic influence

Of the fourteen country domains investigated, six were associated with dominant classifierlanguages as L1s, and eight with countable grammar similar to that employed by ICEnglishes. There is some support in the data for the hypothesis that the rate of pluralizationwould be lower in classifier language domains than in non-classifier language domains:



Figure 3. Countable usage by Kachruvian Circle, ordered by OC, lowest to highest.

between them, the classifier language domains (Korea, Thailand, Bangladesh, Hong Kong,the Philippines and China) exhibit only 1.83 per cent of plural usage, compared with 2.45per cent over all 14 domains (when we exclude the Philippine outliers). Domains withdominant count grammar languages show an average of 2.3 per cent pluralized (thereforecountable) usage. These differences are suggestive, but are not significant statistically andare probably too small to warrant claims for a generalized substrate role.

In order to detect the potential role of cross-linguistic influence from particular nounsin domains where dominant L1s use countable grammar, we obtained translations of themass noun sample set from linguists who were native speakers of, respectively, Sinhala(the dominant language of Sri Lanka), Spanish (dominant in both Mexico and Belize), andSwahili (a Bantu language serving as a national language and lingua franca in Kenya).Informants were asked to provide the first form that occurred to them for each noun in themass-oriented frame ‘A lot of ___’. They were asked also to record any alternative formthat occurred to them as a natural translation.

There is some evidence of a role for cross-linguistic influence, but it is not generalizedand does not appear to be a powerful predictor of countability patterns. In Sinhala, 18 of the25 nouns were expressed by our informant using pluralia tantum; all second preferencesfor translation equivalents, where given, were also plurally inflected nouns. Six nounswere expressed by singular nouns in both first and (where given) second options. Onenoun (traffic) was expressed as first preference by a plural noun phrase or optionally bya singular noun. The average rate of countable use for the nouns pluralized in Sinhalawas 4.98 per cent (4.75% if we include traffic), whereas for the singular nouns it washalf as much, at 2.48 per cent. But the difference is not statistically significant. For theprimarily Spanish-speaking domains, Belize and Mexico, the pattern is repeated, althoughin attenuated form, as we had hypothesized, since unlike the case of pluralia tantum of



Sinhala, Spanish does have singular forms available for all the nouns given as translations.Excluding cases where there is no Spanish translation and the English form has beenborrowed (software and slang), we find a mean of 2.22 per cent countable usage forthe seven nouns translated as plural in Spanish, compared with 1.62 per cent for the 17translated as singular. The difference is not statistically significant.

In the case of domains where Bantu languages are predominantly spoken, namely, Kenyaand Angola, the pattern is murkier. There is little consistency across the figures for the twodomains. In line with our observation regarding OC vs. EC trends, Kenya has over twice therate of countable usage than Angola. But it is not consistently the same nouns which receiveproportionately greater pluralization than IC norms. For example, evidence was translatedby our Swahili-speaking informant as ushahidi, a noun marked with the singular prefixu- (prefix class 11), used more for non-count nouns (Contini-Morava 2000), and althoughonly 0.77 per cent of tokens are pluralized in the Kenya domain, this figure rises to 8.92per cent in the Angola domain. Conversely, the noun with the highest rate of pluralizationfor Kenya is furniture, at 16.96 per cent, compared with only 0.91 per cent for Angola. TheSwahili translation provided by our informant, samani, belongs to the unprefixed class9/10, used mostly with countable nouns. However, a highly pluralized mass noun in theKenya sample is equipment, at 11.29 per cent, expressed with the Swahili noun vifaa,exhibiting the plural prefix vi- (class 8). Equipment is the second highest pluralized formalso for Angola, at 6.5 per cent. The evidence from Swahili noun classes is potentiallyunreliable, however, since the extent to which number-marking patterns are generalizableacross Bantu languages is unclear (Contini-Morava, personal communication). So whilecross-linguistic influence may have a potential role in Englishes with Bantu substrates, itis patently not a defining role.

On the whole, our data suggest that although cross-linguistic influence may explain partof the variability in usage (especially in the case of classifier languages and L1 pluraliatantum), the scepticism of Platt et al. (1984) regarding the importance of substrate influenceis warranted.

Summary

In sum, although these data do not yield much support for the systematic nativizationof mass nouns where the substrate might lead one to expect this, there is some compellingevidence for the greater independence of OC users from IC norms, with over twice therate of plural use attested in ‘second language’ (OC) users than in ‘foreign language’ (EC)users. This difference suggests that it is exposure to, and attitudes to, different models ofEnglish that determine differences in the amount of countable usage of mass nouns, ratherthan the initial L1 state of users’ linguistic knowledge: EC learners tend to converge on ICnorms more than OC learners do, independently of the way their first languages expressthe individuation of nominalizable concepts.

CONCLUSION

The picture that emerges from our data and analysis is one of notable and widespread useof mass nouns used countably, although the phenomenon is highly infrequent in comparisonwith usage that coincides with IC norms. There is little evidence from any domain thatcountable usage is emerging as a new norm across or within OC or EC varieties, or isthe more common of the two grammatical options for these nouns in ELF interaction. Ingeneral, our data do not confirm the suggestion in much of the World Englishes literature



that countable usage of mass nouns is a frequently encountered feature of non-IC Englishes,at least inasmuch as these are represented in the speech of (mostly European) ELF andtext on OC and EC websites. Of course, higher concentrations of countable usage maybe revealed in some contexts through inspection of samples of localized, informal spokeninteraction, especially in African and Asian OC and EC contexts, or in non-Europeanregional ELF (e.g. Kirkpatrick 2010). And particular nouns certainly seem to be moreprone than others to appear countably.

Although our analysis of potential causes yields evidence for a heightened degree of in-novation in OC over EC Englishes, it does not point to a major role for substrate influenceon the basis of lexico-grammatical differences in speakers’ first languages. Claims for‘nativization’ through cross-linguistic influence from L1s are therefore not substantiatedby our data, although there are signs of the greater ‘freedom’ that OC users inherit toconstruct their own mental grammars on the basis of the local and distal models available.What is most surprising about our data is that mass nouns used countably were so hardto find, given their prominence in scholarly and professional work in world Englishes,ELF, and TESOL. It seems that the disproportionate invocation of the count/non-countdistinction in prescriptive works for learners is not because of any major significance it haswithin the overall scheme of English forms and their functions, but rather because of itsstatus as a conspicuous social marker of the border between IC and non-IC Englishes: of‘nativeness’. That is, it is being viewed as part of what Widdowson (1994: 381) identifiedas ‘the symbolic possession of a particular community, expressive of its identity, its con-ventions, and values’. This is recognized in some pedagogically-oriented discussions ofthe phenomenon. In his grammar for English language professionals, for example, Jacobs(1995) observes that ‘errors’ such as many waters

do not hinder communication and are not associated with problems in learners’ comprehension.Nevertheless, frequent determiner errors do pose questions of English language mastery that may disturbemployers, admissions personnel, and teachers of other courses. (Jacobs 1995: 107)

Similarly, McKay (2002) acknowledges that

the pluralization of nouns like equipments and evidences will not cause problems of intelligibility;however, some may contend that such differences reflect a lack of competence on the part of the speakerand be an indication of the deterioration of the language. (McKay 2002: 127)

Clearly, then, monolithic conceptions of English underlie the IC gate-keeping functionsof much TESOL practice, leading to the unquestioned belief that there is a fixed class ofuncountable mass nouns that must be mastered in order to ‘join the club’. Less obviously,however, the highlighting of the phenomenon in descriptions of OC and EC Englishes,often apparently on the basis of casual observation, also invites (neo)-monolithic thinkingabout English (cf. Pennycook 2009: 200). Our low counts in VOICE and the web leadto the conclusion that it has been given prominence because it is typically absent in ICEnglishes rather than because it is present to any great extent outside the IC. Overall, wedo not detect rates of usage that would justify the designation of countable mass nouns asa ‘norm’ for any non-IC national, international or supranational variety.

We suggest that linguistic descriptions which highlight such peripheral and communica-tively inconsequential formal elements on the basis of their contrastiveness with IC Englishrather than quantitative data about actual usage, obscure the plurilithic reality of English



grammars and exaggerate unhelpful monolithic notions of nativeness and non-nativeness.Furthermore, we argue that pedagogical models which forefront such elements in teachingand testing may be doing a disservice to many learners: not because the countability ofmass nouns may be deemed ‘legitimate’ in English models beyond the IC, but because: (a)there is no evidence that it affects communication, especially in ELF contexts; and (b) itperpetuates deficit models of learning by setting up monolithic, purely formal, indicatorsof accomplishment (Hall, forthcoming).

We believe that corpus studies which mine authentic lingua franca usage across a rangeof genres can provide an informative empirical resource with which to assess the waysin which individual mental grammars can coalesce or diverge, and therefore the extent towhich patterns of usage can be alleged to characterize the linguistic resources of somegroups and not others, and so, in turn, be identified as localized or globalized norms.The case discussed here, a lexico-grammatical feature homogenized as a norm for ICEnglishes in the relatively recent past (Crystal 2008; Toyota 2009), has been revealed asa marginal phenomenon in OC and EC Englishes: a shibboleth of the native/non-nativedichotomy rather than a significant feature of specific groups of speakers or situations ofuse. Our analysis demonstrates an approach to the lexico-grammatical forms of Englishwhich stresses their variable and contingent nature, as part of plurilithic systems which arefashioned and refashioned through the communicative practices of individuals and groups(Makoni and Pennycook 2008) and in the minds of millions of individual learners andusers (Hall, forthcoming). Such an approach is consistent with a descriptive linguisticswhich is blind to ‘nativeness’ and an applied linguistics which recognizes the subservienceof form to function.

NOTES

1. The research for this paper was supported by grants to the first author by York St John University Research Officeand HEFCE (TESS funding). We are grateful to Rachel Wicaksono, Nathan Page and anonymous reviewers for theirvery helpful comments on earlier drafts.

2. BNC cannot a priori identify and exclude non-IC Englishes used in British samples, so not all usage in BNC can beassumed to be produced by IC users.

3. Due to the high number of tokens involved, we were unable to inspect surrounding context. A consequence of thiswas that searches for a prior indefinite article were not conducted, as they were with VOICE, because an article +noun sequence can always be the beginning of a noun phrase headed by a compound noun, e.g. a corruption trial ora luggage tag.

4. This filtering process still resulted in data for all 25 mass nouns in all domains, with only two exceptions: Yemen,with 10 nouns eliminated, and Angola, with two.

REFERENCES

Ahulu, Samuel (1998) Grammatical variation in international English. English Today 14, 19–25.Aikhenvald, Alexandra Y. (2003) Classifiers: A Typology of Noun Categorization Devices. Oxford: Oxford University

Press.Allan, Keith (1980) Nouns and countability. Language 56, 541–67.Bamgbos.e, Ayo. (1998) Torn between the norms: Innovations in world Englishes. World Englishes 17, 1–14.Barner, David, Inagaki, Shunji, and Li, Peggy (2009) Language, thought, and real nouns. Cognition 111, 329–44.BBC World Service (2004) Learning English. Virtual English masterclass. Retrieved 6 June 2011 from http://www.

bbc.co.uk/worldservice/learningenglish/events/grammar_transcript.shtml.Bergh, Gunnar (2005) Min(d)ing English language data on the web: What can Google tell us? ICAME Journal 29, 25–46.Bjorkman, Beyza (2008) English as the lingua franca of engineering: The morphosyntax of academic speech events.

Nordic Journal of English Studies 7, 103–22.Bokamba, Eyamba G. (1992) The Africanization of English. In Braj B. Kachru (ed.), The Other Tongue: English across

Cultures (2nd. edn., pp. 125–47). Urbana, IL: University of Illinois Press.



Brook-Hart, Guy (2009) Learning from Common Mistakes. Intermediate. Cambridge: Cambridge University Press.Carter, Ronald, and McCarthy, Michael (2006) Cambridge Grammar of English. Cambridge: Cambridge University

Press.Chierchia, Gennaro (1998) Plurality of mass nouns and the notion of semantic parameter. In Susan Rothstein (ed.), Events

and Grammar. Dordrecht: Kluwer.Contini-Morava, Ellen (2000) Noun class as number in Swahili. In Ellen Contini-Morava and Yishai Tobin (eds.), Between

Grammar and Lexicon (pp. 3–30). Amsterdam: John Benjamins.Crystal, David (2008) Two thousand million? Updates on the statistics of English. English Today 93, 3–6.DeCapua, Andrea (2008) Grammar for Teachers: A Guide to American English for Native and Non-native Speakers. New

York: Springer.Fletcher, William H. (2007) Concordancing the web: Promise and problems, tools and techniques. In Marianne Hundt,

Nadja Nesselhauf and Carolin Biewer (eds), Corpus Linguistics and the Web (pp. 25–46). Amsterdam: Rodopi.Government of Hong Kong SAR (2007) Table 140: Population aged 5 and over by usual language, 1996,

2001 and 2006. Retrieved 19 August 2010 from http://www.censtatd.gov.hk/hong_kong_statistics/statistical_tables/index.jsp?tableID=140&ID=&subjectID=1.

Gramley, Stephan (2001) The Vocabulary of World English. London: Arnold.Gupta, Anthea Fraser (2006) Standard English in the world. In Rani Rubdy and Mario Saraceni (eds), English in the

World: Global Rules, Global Roles (pp. 95–109). London: Continuum.Hall, Christopher J. (forthcoming) Cognitive contributions to plurilithic views of English and other languages. Applied

Linguistics. doi: 10.1093/applin/ams042.Higgins, Christina (2003) “Ownership” of English in the Outer Circle: An alternative to the NS-NNS dichotomy. TESOL

Quarterly 37, 615–44.ICE (2010) International Corpus of English. Retrieved 10 June 2011 from http://ice-corpora.net/ice/index.htmInternet World Stats (2010) Internet world users by language. Retrieved 18May 2010 from http://www.internetworldstats.

com/stats7.htm.Jacobs, Roderick A. (1995) English Syntax. A Grammar for English Language Professionals. Oxford: Oxford University

Press.Jacquemet, Marco (2005) Transidiomatic practices, language and power in the age of globalization. Language and

Communication 25, 257–77.Jenkins, Jennifer (2006) The spread of EIL: A testing time for testers. ELT Journal 60, 42–50.Jenkins, Jennifer (2009) English as a lingua franca: Interpretations and attitudes. World Englishes 28, 200–7.Kachru, Braj B. (1985) Standards, codification and sociolinguistic realism: The English language in the outer circle.

In Randolph Quirk and Henry G. Widdowson (eds), English in the World: Teaching and Learning the Language andLiteratures (pp. 11–30). Cambridge: Cambridge University Press.

Kachru, Braj B. (1992) Models for non-native Englishes. In Braj B. Kachru (ed.), The Other Tongue: English acrossCultures (2nd edn., pp. 48–74). Urbana, IL: University of Illinois Press.

Kachru, Yamuna, and Smith, Larry (2008) Cultures, Contexts, and World Englishes. New York: Routledge.Keller, Frank, and Lapata, Mirella (2003) Using the web to obtain frequencies for unseen bigrams. Computational

Linguistics 29, 459–84.Kirkpatrick, Andy (2010) Researching English as a lingua franca in Asia: The Asian Corpus of English (ACE) project.

Asian Englishes 13, 4–18.Kirkpatrick, Andy, and Deterding, David (2011) World Englishes. In James Simpson (ed.), The Routledge Handbook of

Applied Linguistics (pp. 373–87). London: Routledge.Lapata, Mirella, and Keller, Frank (2005) Web-based models for natural language processing. ACM Transactions on

Speech and Language Processing 2, 1–30.Lewis, M. Paul (ed.) (2009) Ethnologue: Languages of the World (16th edn.). Dallas, TX: SIL International.Lowenberg, Peter H. (1992) Testing English as a world language: Issues in assessing non-native proficiency. In Braj B.

Kachru (ed.), The Other Tongue. English across Cultures (2nd edn., pp. 108–21). Urbana, IL: University of IllinoisPress.

Makoni, Sinfree, and Pennycook, Alistair (2007) Disinventing and reconstituting languages. In Sinfree Makoni andAlistair Pennycook (eds.), Disinventing and Reconstituting Languages (pp. 1–41). Clevedon: Multilingual Matters.

McArthur, Tom (2002) The Oxford Guide to World English. Oxford: Oxford University Press.McKay, Sandra L. (2002) Teaching English as an International Language: Rethinking Goals and Approaches. Oxford:

Oxford University Press.Mesthrie, Rajend, and Bhatt, Rakesh M. (2008) World Englishes. The Study of New Linguistic Varieties. Cambridge:

Cambridge University Press.Mesthrie, Rajend, and West, Paula (1995) Towards a grammar of proto South African English. English World-Wide 16,

105–34.Mollin, Sandra (2007) New variety or learner English? Criteria for variety status and the case of Euro-English. English

World-Wide 28, 167–85.Norvig, Peter (2007) Theorizing from data: Avoiding the capital mistake. Video lecture. Retrieved 18 May 2010 from

http://www.youtube.com/watch?v=nU8DcBF-qo4.



Pennycook, Alistair (2009) Plurilithic Englishes: Towards a 3D model. In Kumiko Murata and Jennifer Jenkins (eds),Global Englishes in Asian Contexts. Current and Future Debates (pp. 194–207). Basingstoke: Palgrave Macmillan.

Platt, John T., Weber, Heidi and Ho, Mian Lian (1984) The New Englishes. London: Routledge and Kegan Paul.Quirk, Randolph (1990) Language varieties and standard language. English Today 6, 3–10.Quirk, Randolph, Greenbaum, Sidney, Leech, Geoffrey, and Svartvik, Jan (1985) A Comprehensive Grammar of the

English Language. London: Longman.Rajagopalan, Kanavillil (1997) Linguistics and the myth of nativity: Comments on the controversy over ‘new/non-native’

Englishes, Journal of Pragmatics 27, 225–31.Renouf, Antoinette (2003) WebCorp: Providing a renewable data source for corpus linguists. In Sylviane Granger and

Stephanie Petch-Tyson (eds.), Extending the Scope of Corpus-based Research: New Applications, New Challenges.Amsterdam: Rodopi.

Schmied, Josef (2006) East African Englishes. In Braj B. Kachru, Yamuna Kachru and Cecil L. Nelson (eds), TheHandbook of World Englishes (pp. 188–202). Oxford: Blackwell.

Schneider, Edgar W. (2011) English Around the World. Cambridge: Cambridge University Press.Schoenberg, Irene E. (2005) Focus on Grammar: A Basic Course for Reference and Practice. London: Longman.Seidlhofer, Barbara (2005) English as a lingua franca. In Albert S. Hornby (ed.), Oxford Advanced Learner’s Dictionary

of Current English (7th edn.), Oxford: Oxford University Press.Seidlhofer, Barbara (2009) Orientations in ELF research: Form and function. In Anna Mauranen and Elina Ranta (eds),

English as a Lingua Franca: Studies and Findings (pp. 37–59). Newcastle upon Tyne: Cambridge Scholars.Sey, Kofi A. (1973) Ghanaian English: An Exploratory Survey. London: Macmillan.Shim, Rosa J. (1999) Codified Korean English: Process, characteristics and consequence. World Englishes 18, 247–58.Slobin, Dan (1973) Cognitive prerequisites for the development of grammar. In Charles Ferguson and Dan Slobin (eds),

Studies in Child Language Development (pp. 175–208). New York: Holt, Rinehart and Winston.Swan, Michael (1995) Practical English Usage. Oxford: Oxford University Press.Toyota, Junichi (2009) When the mass was counted: English as classifier and non-classifier language. SKASE Journal of

Theoretical Linguistics 6, 118–30.Villavicencio, Aline (2003) Verb-particle constructions in the world wide web. Proceedings of the ACL-SIGSEM Work-

shop on the Linguistic Dimensions of Prepositions and their Use in Computational Linguistics Formalisms andApplications. Toulouse, France. Retrieved 18 May 2010 from http://www.cl.cam.ac.uk/∼av208/vpc-www.final.ps.gz.

VOICE (2009) The Vienna-Oxford International Corpus of English. Retrieved 11 October 2009 from http://voice.univie.ac.at.

Widdowson, Henry G. (1994) The ownership of English. TESOL Quarterly 28, 377–89.Williams, J. (1987) Non-native varieties of English: A special case of language acquisition. English World-Wide 8, 161–99.Xu, Fei (2007) Concept formation and language development: Count nouns and object kinds. In Gareth Gaskell (ed.),

The Oxford Handbook of Psycholinguistics (pp. 627–34). Oxford: Oxford University Press.Yates, Jean (2006) English Vocabulary for Beginning ESL Learners. New York: McGraw-Hill.

(Received 13 June 2012)


Documents

Countability in world Englishes...context of TESOL and the descriptive contexts of world Englishes and English as a Lingua Franca. We advocate the use of web-based corpora to investigate